摘要統計

Python 的 Monte Carlo 模擬

Izzy Weber

Curriculum Manager, DataCamp

年齡與結果

第一與第四個年齡分位的族群，結果差多少？模擬結果中 Age 變數的長條圖

年齡與結果

print(np.quantile(df_summary["age"], 0.25))
print(np.quantile(df_summary["age"], 0.75))

39.78584463094688
57.52092642086441

年齡與結果

age_q25 = np.quantile(df_summary["age"], 0.25)
age_q75 = np.quantile(df_summary["age"], 0.75)


mean_age_q75_outcome = np.mean(df_summary[df_summary["age"] > age_q75]["predicted_y"]) 
mean_age_q25_outcome = np.mean(df_summary[df_summary["age"] < age_q25]["predicted_y"])


mean_age_q75_outcome - mean_age_q25_outcome

34.09429663553621

依年齡與 bmi 的結果差異

bmi 與年齡第 1 與第 4 分位的族群，結果差多少？
這個差異的 95% 信賴區間與標準差是多少？

依年齡與 bmi 的結果差異

y_diffs = []
for i in range(1000):
    simulation_results = st.multivariate_normal.rvs(mean=mean_dia, size=1000, cov=cov_dia)
    df_results = pd.DataFrame(simulation_results, columns=["age", "bmi", "bp", "tc", "ldl", 
                                                           "hdl", "tch", "ltg", "glu"])

    predicted_y = regr_model.predict(df_results)
    df_y = pd.DataFrame(predicted_y, columns=["predicted_y"])
    df_sum = pd.concat([df_results, df_y], axis=1)

    age_q25 = np.quantile(df_sum["age"], 0.25)
    age_q75 = np.quantile(df_sum["age"], 0.75)
    bmi_q25 = np.quantile(df_sum["bmi"], 0.25)
    bmi_q75 = np.quantile(df_sum["bmi"], 0.75)

    q75_outcome = np.mean(df_sum[(df_sum["bmi"] > bmi_q75) & (df_sum["age"] > age_q75)]["predicted_y"]) 
    q25_outcome = np.mean(df_sum[(df_sum["bmi"] < bmi_q25) & (df_sum["age"] < age_q25)]["predicted_y"])

    y_diff = q75_outcome - q25_outcome 
    y_diffs.append(y_diff)

依年齡與 bmi 的結果差異

平均值：
```
np.mean(y_diffs)
```

132.4948511247819

95% 信賴區間：
```
np.quantile(y_diffs,0.025)
```

120.73340800299707

np.quantile(y_diffs,0.975)

144.1344994507557

標準差：
```
np.std(y_diffs)
```

5.9322225537128

結果的長條圖：
```
sns.histplot(y_diffs)
```

一起來練習吧！

Python 的 Monte Carlo 模擬